Discovery of the Membrane Binding Domain in Trifunctional ProlineUtilization AShelbi L. Christgen,§,‡ Weidong Zhu,§,‡ Nikhilesh Sanyal,§ Bushra Bibi,§ John J. Tanner,∥,⊥

and Donald F. Becker*,§

§Department of Biochemistry, Redox Biology Center, University of Nebraska-Lincoln, Lincoln, Nebraska 68588, United States∥Departments of Biochemistry and ⊥Chemistry, University of Missouri-Columbia, Columbia, Missouri 65211, United States

*S Supporting Information

ABSTRACT: Escherichia coli proline utilization A (EcPutA) is the archetype oftrifunctional PutA flavoproteins, which function both as regulators of the prolineutilization operon and bifunctional enzymes that catalyze the four-electron oxidationof proline to glutamate. EcPutA shifts from a self-regulating transcriptional repressorto a bifunctional enzyme in a process known as functional switching. The flavinredox state dictates the function of EcPutA. Upon proline oxidation, the flavinbecomes reduced, triggering a conformational change that causes EcPutA todissociate from the put regulon and bind to the cellular membrane. Major structure/function domains of EcPutA have been characterized, including the DNA-bindingdomain, proline dehydrogenase (PRODH) and L-glutamate-γ-semialdehydedehydrogenase catalytic domains, and an aldehyde dehydrogenase superfamilyfold domain. Still lacking is an understanding of the membrane-binding domain,which is essential for EcPutA catalytic turnover and functional switching. Here, weprovide evidence for a conserved C-terminal motif (CCM) in EcPutA having a critical role in membrane binding. Deletion of theCCM or replacement of hydrophobic residues with negatively charged residues within the CCM impairs EcPutA functional andphysical membrane association. Furthermore, cell-based transcription assays and limited proteolysis indicate that the CCM isessential for functional switching. Using fluorescence resonance energy transfer involving dansyl-labeled liposomes, residues inthe α-domain are also implicated in membrane binding. Taken together, these experiments suggest that the CCM and α-domainconverge to form a membrane-binding interface near the PRODH domain. The discovery of the membrane-binding region willassist efforts to define flavin redox signaling pathways responsible for EcPutA functional switching.

Proline and its metabolism have numerous significant roles ina myriad of cellular processes.1−4 Cellular energy is

generated from the oxidative breakdown of proline to glutamatevia the enzymes proline dehydrogenase (PRODH) and L-glutamate-γ-semialdehyde dehydrogenase (GSALDH), which ineukaryotes are localized in the mitochondrion.2,5,6 In humans,PRODH has been shown to influence cancer development andprogression, with suppressor and pro-metastasis functions inearly and late stage cancer, respectively.2,4,7−10 Inborndeficiencies in PRODH are associated with hyperprolineamiatype 1,11 risk of schizophrenia,11−13 and autism spectrumdisorder.14 The role of proline metabolism in infectious diseasesis of significant interest as proline has been shown to supportgrowth and virulence of different pathogenic organisms includingStaphylococcus aureus,15,16 Helicobacter pylori,17,18 Cryptococcusneoformans,19 and trypanosomes.20−22 The primary functions ofproline during infection include utilization as an energy, carbon,and nitrogen source, as well as osmoregulation and stressresistance.In Gram-negative bacteria, the oxidation of proline to

glutamate is catalyzed by the enzyme proline utilization A(PutA), which has PRODH and GSALDH domains in onepolypeptide.5,23 PutA enzymes are unique to Gram-negative

bacteria, as eukaryotes and Gram-positive bacteria encodeseparate PRODH and GSALDH enzymes.5 The oxidation ofproline to glutamate is initiated by PRODH, which generatesΔ1-pyrroline-5-carboxylate (P5C) (Figure 1A). PRODH is a FAD-dependent (βα)8 barrel domain enzyme that couples theoxidation of proline to reduction of ubiquinone in the respiratorychain.5,24−26 GSALDH (aka P5C dehydrogenase andALDH4A1) catalyzes the NAD+-dependent oxidation of L-glutamate-γ-semialdehyde (GSAL), which is generated in anintervening nonenzymatic step by hydrolysis of P5C.5,27 Asubstrate channel links the PRODH and GSALDH domains inPutAs, thereby preventing P5C/GSAL loss to the cellular milieuand improving efficiency of glutamate production from pro-line.3,5,27−32 Recent evidence suggests that substrate channelingalso occurs between separate PRODH and GSALDH enzymesfrom Gram-positive Thermus thermophilus, indicating thatsubstrate channeling may be a fundamental feature of prolinemetabolism in other organisms that lack PutA.33

Received: October 5, 2017Revised: October 26, 2017Published: November 1, 2017

Article

pubs.acs.org/biochemistry

© 2017 American Chemical Society 6292 DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

Cite This: Biochemistry 2017, 56, 6292-6303

PutA proteins are classified into three main types dependingon the domain architecture (Figure 1B). PutAs in the type A classcontain the two catalytic domains and are the best structurallycharacterized with high-resolution crystal structures of PutAfrom Bradyrhizobium japonicum (BjPutA),27 Geobacter sulfurre-ducens (GsPutA),28 and Bdellovibrio bacteriovorus.34 Type BPutAs are distinguished by an additional C-terminal domain thatwas recently identified as an aldehyde dehydrogenase super-family (ALDHSF) fold. Characterization of type B PutAs fromSinorhizobium meliloti29 and Corynebacteirum f reiburgense35

showed that the major function of the C-terminal ALDHSFdomain is to stabilize aldehyde binding in the GSALDH activesite and seal the substrate channel from bulk solvent.5,36 Type APutAs, which lack the ALDHSF domain, have a differentoligomeric structure than type B PutAs and form the substratechannel via domain-swapped dimerization.5 Finally, a group ofPutA proteins classified as type C contain a N-terminal DNA-binding domain and are known as “trifunctional PutAs” (Figure1B).5 The DNA-binding domain has a ribbon-helix-helix (RHH)fold and has been structurally characterized from Escherichiacoli37,38 and Pseudomonas putida.39

Trifunctional PutAs repress transcription of the putA and putP(encoding a high affinity Na+-proline symporter)40 genes of theput regulon in response to proline availability.37,41−43 As prolinelevels rise, PutA switches from a DNA-bound transcriptionalrepressor to a membrane-bound catabolic enzyme therebyrelieving repression of the put regulon and increasing thebreakdown of proline.44 Previous work with PutA from E. coli(EcPutA) has shown proline reduction of the FAD cofactortriggers a conformational change that significantly enhancesPutA−membrane associations effectively transforming PutAfrom a soluble DNA-binding protein to a membrane-bound

enzyme.45−47 Reduction of the FAD cofactor alone is sufficientto induce conformational changes and membrane binding.45 X-ray crystal structures of the EcPutA PRODH domain in differentredox states and studies of PRODH active site mutants implicatea hydrogen bond network from the FAD N(5) position to theβ3-α3 loop in the PRODH domain as critical for EcPutAfunctional switching.48−50 Mutation of hydrogen bondingresidues in the network traps EcPutA as a transcriptionalrepressor with the mutants unable to localize to the membraneupon proline reduction of the FAD cofactor.48,50

While significant progress has been made identifying residuescritical for EcPutA functional switching, a major gap exists inunderstanding how EcPutA associates with membranes. Priorstudies, however, have provided some important clues. Crystalstructures of PutAs show three conserved ancillary domains(arm, α, and linker domains) near the PRODH catalyticdomain.5,27−29,51 The proximity of the arm and α domains tothe PRODH active site suggests they are positioned to respondto flavin reduction. In fact, proline reduction of EcPutA increasessusceptibility to chymotrypsin cleavage at Arg234, implying thatthe surrounding region within the α-domain (residues 142−259)becomes more solvent exposed with flavin in the reducedstate.46,52 It was also shown that loss of residues 58−233eliminates EcPutA functional membrane associations whilehaving no impact on PRODH activity.46 Recently, a crystalstructure of GsPutA showed a detergent molecule bound in theα-domain, supporting the idea of this domain being involved inmembrane binding.28 Although GsPutA is classified as a shortbifunctional PutA (i.e., type A), it shares high structuralhomology with the EcPRODH domain, indicating a likelyconserved function for the α-domain.28

Figure 1. (A) Reactions catalyzed by PRODH and GSALDH resulting in the oxidation of proline to glutamate. (B) Domain architectures of PutAs.EcPutA (1320 amino acids) is classified as type C. (C) Results from a multiple sequence alignment of PutAs showing the conservation of residues in theCCM. The alignment was computed using the T-Coffee server and visualized using WebLogo software. Residue numbering is according to the EcPutAsequence. The GenBank accession numbers of the aligned PutA sequences and T-Coffee alignment of these residues are provided in Table S2,Supporting Information.

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6293

Another region implicated in EcPutA−membrane bindingincludes residues in the C-terminal end. A previous study showedthat deletion of EcPutA residues 1295−1320 (EcPutA1−1294mutant) generates a constitutive transcriptional repressor(EcPutA1−1294 mutant) that is devoid of functional membraneassociations indicating an important role for C-terminal residuesin EcPutA−membrane binding and functional switching.44

To further understand how EcPutA switches function, wesought to identify the membrane-binding domain(s) of PutA.Here, we discover the essential function of a conserved C-terminal motif (CCM, residues 1300−1320, Figure 1C) in PutAmembrane binding and identify residues that are critical forEcPutA functional switching. We also show evidence for the α-domain having an important role in PutA−membrane binding.Despite the large separation in primary sequence, we proposethat residues of the CCM and α-domain converge spatially in thestructure of PutAs to generate a membrane-binding interface.

■ EXPERIMENTAL PROCEDURESMaterials. All chemicals and buffers were purchased from

Fisher Scientific and Sigma-Aldrich, unless otherwise noted. E.coli strains XL-Blue and BL21(DE3) pLysS were purchased fromNovagen and used for cloning and protein expression,respectively.An E. coli BL21(DE3) strain that lacks PutA was used to test in

vivomembrane interactions of PutA. A PutA deficient (putA−) E.coli strain of BL21(DE3) was generated by disrupting the putAgene with a kanamycin cassette (TargeTron Gene knockoutsystem, Sigma). The primers used were 5′-AAAAAA-GCTTATAATTATCCTTACGCCTCCCGCAGGTGC-GCCCAGATAGGGTG-3′, 5′-CAGATTGTACAAATG-TGGTGATAACAGATAAGTCCCGCAGCCTAA -C T T A C C T T T C T T T G T - 3 ′ a n d 5 ′ -TGAACGCAAGTTTCTAATTTCGGTTAGGCGTC-GATAGAGGAAAGTGTCT-3′. The kanamycin cassette wasinserted at 355 bp into the putA gene. Subsequent PCR analysisof the putA gene showed a 547 bp product, which contains 219bp of the kanamycin cassette, demonstrating insertion of thekanamycin cassette into the putA gene. Western blot analysis ofcell lysates using an antibody generated against the RHH domainas previously described44 confirmed the lack of PutA expression.Mutagenesis and Protein Purification. Site-directed

mutants of EcPutA were made from the full-length construct ofwild-type EcPutA (UniProt Accession ID:P09546) (EcPutA-pET14b) using Stratagene QuikChange or Invitrogen GeneTai-lor mutagenesis kits. The primers used for mutagenesis werepurchased from Integrated DNA technologies and are listed inTable S1 (see Supporting Information). C-terminal deletionmutants were generated by inserting a stop codon immediatelyafter the codons for residues 1308, 1314, and 1317. All of theEcPutA mutants were confirmed by DNA sequencing. EcPutAwild-type andmutant proteins were overexpressed in E. coli strainBL21(DE3) pLysS and purified by Ni-NTA Superflow affinity(Qiagen) chromatography using a N-terminal 6xHis-tag aspreviously described.50,53 The N-terminal His-tag was retainedfor EcPutA wild-type and mutants in subsequent experiments.Purified proteins were stored in 50 mMTris buffer containing 50mM NaCl, 0.5 mM EDTA, 0.5 mM Tris (3-hydroxypropyl)phosphine (THP), and 10% glycerol (pH 7.5).PRODH Kinetic Assays. The PRODH kinetic parameters of

EcPutA wild-type and mutants were determined usingdichlorophenolindophenol (DCPIP) as the terminal electronacceptor and phenazine methosulfate as the mediator

(proline:DCPIP oxidoreductase assay) as previously described.36

The steady-state parameters Km and kcat for proline weredetermined as previously described using EcPutA protein (0.01−0.3 μM) and L-proline (0−700 mM) while keeping DCPIP fixedat 75 μM. Assays were performed in 20 mMTris-HCl buffer (pH8.0) with 10% glycerol at 23 °C.

Limited Proteolysis of EcPutA Proteins. Limiteddigestion of EcPutA wild-type and mutants by chymotrypsinwas performed as previously described.49 Briefly, digests wereperformed with and without 20 mM proline in 50 mMpotassiumphosphate buffer (pH 7.5, 10% glycerol) for 1 h at 23 °C. Afterstopping the reactions with phenylmethanesulfonyl fluoride andhot SDS-PAGE sample buffer, samples were analyzed by SDS-PAGE (8% acrylamide). Resulting polypeptide bands werevisualized with Coomassie Blue G-250 staining.

Functional Membrane Association and Lipid Pull-Down Assays. Functional membrane association activity ofwild-type EcPutA and the EcPutA variants was measured aspreviously described by monitoring the formation of a P5C/o-aminobenzaldehyde (o-AB) yellow adduct at 443 nm (ε = 2590M−1 cm−1) in a 96-well plate.48 Assays (200 μL total volume)were performed at 23 °C in 50 mM 3-(N-morpholino)-propanesulfonic acid (MOPS) buffer (10 mM MgCl2, 10%glycerol, pH 7.4) with 4 mM o-AB, 60 mM L-proline, and 0.02mg/mL membrane vesicles prepared from E. coli strain JT31putA−. Inverted E. coli membrane vesicles were prepared asreported by Abrahamson et al., frozen in liquid N2, and stored at−70 °C until needed.54

Lipid pull-down assays were performed as previouslydescribed.48,49 Assays were performed anaerobically under anitrogen atmosphere in an anaerobic chamber (Belle Technol-ogy Glovebox) with vesicles prepared from E. coli polar lipids(Avanti Polar Lipids).48,49 PutA proteins were incubated with E.coli polar lipid vesicles (0.8 mg/mL) in 10 mMHEPES (pH 7.4)with 150 mMNaCl with or without 20 mM L-proline for 1 h andthen centrifuged.49 The soluble and lipid fractions were thenanalyzed by SDS-PAGE (10% gels) and stained with CoomassieBlue G-250. Partition coefficients for EcPutA wild-type andmutants were calculated as previously described by dividing thesoluble/lipid (S/L) distribution ratio of the oxidized protein (S/L)ox by the distribution ratio of the proline reduced protein (S/L)red.

44 The partition coefficient is thus the quantity (S/L)ox/(S/L)red and indicates the extent to which proline induces PutApartitioning into the lipid fraction.44 Complete reduction of theEcPutA wild-type and mutant proteins by 20 mM L-proline wasconfirmed by following reduction of the FAD cofactor at 450 nmby UV−visible spectroscopy.

Cell-Based Assays.Cell-based transcription repressor assayswere performed as previously described in E. coli strain JT31putA- lacZ-containing the putC:lacZ reporter gene (pUT03construct) and constructs for EcPutA wild-type and mutantsEcPutA1−1314, EcPutA1−1308, L1316D, and A1309D/L1316D (pUC18-EcPutA).48 Cells were grown at 37 °C inminimal media containing ampicillin (50 μg/mL) andchloramphenicol (30 μg/mL) supplemented with 0 and 15mM L-proline and grown to OD at 600 nm ∼1.0. β-galactosidaseassays were performed using cell extracts as describedpreviously48 with β-galactosidase activity reported in Millerunits according to the method of Miller.55

Membrane interactions were tested in vivo using triphenylte-trazolium chloride (TTC) proline indicator plates according tothe method of Wood56 and using the E. coli strain BL21(DE3)Kanr putA−. E. coli cells were transformed with constructs

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6294

EcPutA-pET14b, EcPutA(1−1308)-pET14b, BjPutA-pKA8H,and BjPutA(1−986)-pKA8H.36 To test for functional membraneinteractions, cells were grown in Luria−Bertani (LB) medium toan OD600 of 0.5 and streaked on TTC-indicator agar mediumcontaining L-proline (0.5%), peptone (0.2%), L-tryptophan(0.005%), thiamine (1 μg/mL), TTC (0.0025%) (w/v %),ampicillin (50 μg/mL), and kanamycin (25 μg/mL).Forster Resonance Energy Transfer (FRET) Assays.

Synthetic dansyl-PE (1,2-dioleoyl-sn-glycero-3-phosphoethanol-amine-N-(5-dimethylamino-1-naphthalenesulfonyl) lipids and E.coli polar lipid extract (67% phosphoethanolamine, PE; 23.2%phosphatidylglycerol, PG; and 9.8% cardiolipin) were purchasedfrom Avanti Polar Lipids. Lipids were dried under N2 andresuspended in 10 mM N-(2-hydroxyethyl)piperazine-N′-2-ethanesulfonic acid (HEPES) (pH 7.4) containing 150 mMNaCl and 2 mM MgCl2 (HEPES-NM buffer). The dansyl-PElipid and E. coli polar lipid suspensions were then mixed to makea 5% dansyl-PE and 95% E. coli polar lipid suspensionmixture at afinal lipid concentration of 1 mg/mL. Lipid vesicles of the 5%dansyl-PE/95% E. coli lipid suspension were then prepared aspreviously described by passing the lipids through a 100 nmpolycarbonate filter using a LiposoFast microextruder.57 For thefluorescence measurements, EcPutA wild-type and mutants (1μM) were incubated for 5 min with 50 μg of 5% dansyl-PE/95%E. coli liposomes in HEPES-NM buffer at 23 °C. Fluorescenceemission was measured from 305 to 580 nm using an excitationwavelength of 295 nm (λex 295 nm). L-Proline (5 mM finalconcentration) was then added to the liposome suspension andincubated for 1 min. After incubation with proline, the emissionspectrum was recorded again (λex 295 nm). Normalized percentFRET change was calculated by normalizing the change inemission at 520 nm between 0 and 5 mM proline relative to thefluorescence emission change observed with EcPutA wild-type.Statistical significance was determined using a two-tailed t testusing Sigma Plot 12.0 software for which a p-value <0.05indicated a significant difference in FRET change.

■ RESULTSThe Conserved C-terminal Motif is Essential for

Membrane Binding. Because the EcPutA1−1294 mutant wasshown previously to lack membrane binding, residues 1295−1320 were analyzed further for potential membrane inter-actions.44 In a recent study a conserved C-terminal motif (CCM)was identified and shown to have a role in substrate channeling inPutA from Rhodobacter capsulatus (RcPutA) and Bradyrhizobiumjaponicum (BjPutA).36 We performed a multiple sequencealignment of PutAs (79 total) to investigate the conservationof hydrophobic residues in the CCM of EcPutA, which includesresidues 1300−1320. A stretch of conserved hydrophobicresidues were found in the CCM of EcPutA beginning at residue1308 with the most highly conserved residues corresponding toAla1309 and Leu1316 (Figure 1C). The multiple sequencealignment of the CCM is provided in Figure S1, SupportingInformation.The potential role of the CCM in membrane binding was

examined by generating the CCM deletion mutant, EcPutA1−1308. The ability of EcPutA1−1308 to interact with membraneswas first tested by cell-based assays using TTC-proline indicatormedium.56 E. coli cells deficient in PutA (putA−) containingeither EcPutA wild-type or EcPutA1−1308 constructs weretested for the ability to couple proline utilization with therespiratory chain by following reduction of TTC, a redox dye thatacts as an electron acceptor for dehydrogenases and membrane

respiratory complexes.58,59 Figure 2 shows that cells with wild-type EcPutA turn red indicating TTC reduction, whereas cells

with EcPutA1−1308 are white, suggesting that EcPutA1−1308 isdeficient in membrane binding. To further corroborate the roleof CCM in membrane binding, E. coli putA− cells weretransformed with BjPutA wild-type and the CCM deletionmutant BjPutA1−986. Previously, BjPutA1−986 was shown toexhibit PRODH and GSALDH activity similar to that of wild-type BjPutA.36 However, the membrane binding activity was notexamined. Figure 2 shows that cells with wild-type BjPutAgenerate a red color whereas cells containing BjPutA1−986remain white, further implicating the CCM as critical for PutA-membrane interactions.

Roles of CCM Residues in Functional MembraneAssociations. To further examine the membrane-binding roleof the CCM in EcPutA, two other truncation mutants besidesEcPutA1−1308 were generated. As detailed in Figure 3, thesetruncations included mutants EcPutA1−1314 and EcPutA1−1317. In addition, point mutations were generated by replacinghydrophobic residues (Ala1309 and Leu1316) in the CCMregion with Asp as summarized in Figure 3. The ability of thesemutants to functionally associate with E. coli membrane vesicleswas first determined by PRODH activity assays in whichmembrane vesicles are the terminal electron acceptor. The abilityof EcPutA to couple proline oxidation with reduction ofubiquinone in the membrane during catalytic turnover isindicated by formation of a P5C/o-AB yellow complex, whichis followed at 443 nm. Truncation of C-terminal residues resultedin a trend of decreased P5C/o-AB formation with EcPutA1−1308 showing no measurable activity (Figure 3). In assays usingDCPIP as the terminal electron acceptor, the truncation mutantsexhibited PRODH activity similar to that of wild-type EcPutA(Table 1). These results indicate that the decrease in P5Cproduction with the truncation mutants is due to an attenuated

Figure 2. Cell-based membrane interactions. E. coli strain BL21(DE3)putA− was transformed with wild-type E. coli (EcPutA) and B. japonicum(BjPutA) PutA constructs and conserved C-terminal motif (CCM)deletion mutant EcPutA1−1308 and BjPutA1−986 constructs. Cellswere plated on triphenyltetrazolium chloride (TTC)-indicator agarmedium containing L-proline and grown for 20 h at 37 °C. Red coloniesindicate the ability to utilize proline as a respiratory substrate andfunctional PutA−membrane interactions.

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6295

ability to functionally utilize membrane vesicles as an electronacceptor, rather than a general loss of PRODH activity.Substitution of Ala1309 or Leu1316 with Asp also resulted inlower functional membrane activity (Figure 3B). In theEcPutAA1309D/L1316D double mutant, functional membraneassociation activity was abolished, similar to that observed for thetruncation mutant EcPutA1−1308. Themutants EcPutAL1316Dand EcPutAA1309D/L1316D exhibit PRODH:DCPIP oxidor-eductase activity similar to that of wild-type EcPutA, indicatingthat the loss of activity with membranes is due to dysfunctionalmembrane binding (Table 1).

CCM Alteration Attenuates Physical Membrane Asso-ciation. Next, we sought to determine if the loss of functionalassociation activity results from loss of physical associationsbetween the mutants and membrane vesicles. To this extent, weexamined the ability of the EcPutA mutants to physically interactwith liposomes using lipid pull-down assays in the absence andpresence of proline. Without proline, wild-type EcPutA resides inthe soluble fraction but upon adding proline, EcPutA sedimentswith the liposomes (made from E. coli polar lipids) aftercentrifugation (Figure 4A). The distribution of EcPutA betweenthe lipid and soluble fractions was quantified by densitometryand is plotted in Figure 4B. A partition coefficient was thendetermined as a measure of the EcPutA shift from soluble to lipidfraction upon proline reduction of the flavin.48

Figure 4 shows the results of lipid-pull down assays with thedifferent EcPutA mutants and the corresponding distribution inthe soluble and lipid fractions. The amount of PutA protein in thelipid fraction after adding proline was generally lower for theEcPutA mutants relative to that of wild-type EcPutA. MutantsEcPutA1−1308 and EcPutAA1309D/L1316D exhibited thelowest abundance in the lipid fraction indicating deficient lipidinteractions. These results are consistent with the functionalmembrane association assays in which nomeasurable activity wasobserved with mutants EcPutA1−1308 and A1309D/L1316D.The partition coefficient for EcPutA wild-type is 8.8, while theA1309D/L1316D mutant has the lowest value of 1.2 (Figure4B). A lower partition coefficient value relative to wild-typeEcPutA is consistent with weak proline induction of membranebinding in the mutant A1309D/L1316D. These results furtherdemonstrate that loss of CCM residues in EcPutA, orreplacement of hydrophobic residues with negatively chargedaspartate residues in the CCM, impairs EcPutA membranebinding.To further explore EcPutA membrane interactions, we utilized

a FRET approach with EcPutA intrinsic tryptophan fluorescenceas the donor (emission λmax 348 nm) and fluorescent dansylliposomes as the acceptor (excitation λmax 336 nm, emission λmax520 nm). EcPutA wild-type was first incubated with dansyl-labeled liposomes, and the fluorescence spectrum was recorded.Next, the wild-type EcPutA−liposome mixture was titrated withincreasing amounts of proline (0−10 mM) (Figure 5). Theintrinsic Trp fluorescence decreased as a function of proline,whereas the dansyl-labeled liposomes increased in emission at520 nm. These results are consistent with proline reduction ofthe FAD cofactor triggering EcPutA−membrane binding asindicated by increased FRET between Trp residues (donor) andthe dansyl-labeled liposomes (acceptor). Thus, EcPutA−lipidbinding results in excitation of the liposomes by Trp fluorescenceemission leading to increased dansyl fluorescence of nearly 13%with wild-type EcPutA (λmax 520 nm).FRET experiments were then performed with the EcPutA

mutants EcPutA1−1308, EcPutA1−1314, EcPutA1−1317,

Figure 3. Functional membrane binding analysis of EcPutA C-terminalmutants. (A) Sequences of the CCM deletions in EcPutA truncationmutants. Locations of residues mutated to aspartate (Ala1309 andLeu1317) are indicated in red. (B) Functional membrane associationactivity of EcPutA wild-type and mutants was measured by incubating 4μg of EcPutA protein with 0.1 mg/mL E. coli (strain JT31 putA−)inverted membrane vesicles, 60 mM proline, and 4 mM o-AB in 20 mMMOPS (pH 7.5). All reactions were performed at 23 °C, and absorbancewas monitored at 443 nm, which follows o-AB-P5C adduct formation.

Table 1. PRODH Kinetic Parametersa for EcPutA Wild-Typeand Mutants

EcPutA Km (mM) kcat (s−1) kcat/Km (M−1 s−1)

wild-type 166.7 ± 4.8 10.5 ± 0.1 63.2 ± 1.9EcPutA1−1308 119.6 ± 5.6 12.1 ± 0.2 104 ± 9EcPutA1−1314 171.4 ± 7.9 14.8 ± 0.3 86.5 ± 4.4EcPutA1−1317 238.1 ± 22.4 14.0 ± 0.6 58.8 ± 6.1L1316D 170.6 ± 13.3 18.7 ± 0.6 109.3 ± 9.2A1309D/L1316D 141.8 ± 5.6 10.9 ± 0.3 76.8 ± 3.3W194F 125.6 ± 36.2 6.4 ± 0.5 50.8 ± 18.8W211F 188.0 ± 40.8 6.9 ± 0.5 36.6 ± 10.5W194F/W211F 60.1 ± 12.3 4.9 ± 0.3 80.7 ± 21.3W396F 237.6 ± 65.8 10.9 ± 1.6 46.1 ± 19.5W1185F 92.2 ± 19.4 13.7 ± 1.0 143.9 ± 40.6F259W 103.9 ± 20.6 9.6 ± 0.7 92.6 ± 25.2M255W/F259W 312.9 ± 105 9.9 ± 1.6 31.6 ± 15.7F205W/F259W 220.1 ± 74.2 3.3 ± 0.6 14.9 ± 7.6G1320W 100.4 ± 18.2 7.8 ± 0.5 77.7 ± 19.4L131W 287 ± 10.3 4.9 ± 1.1 17.1 ± 4.5Y129W 157 ± 15.5 17.0 ± 0.1 108.2 ± 11.4

aKinetic parameters were determined in 20 mM Tris-HCl buffer (pH8.0, 10% glycerol) containing 75 μM DCPIP and varying prolineconcentrations at 23 °C.

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6296

L1316D, and A1309D/L1316D. Figure 5B shows the prolinedependent increase in FRET relative to that observed with wild-type EcPutA. EcPutA1−1317 showed FRET similar to that ofwild-type, whereas with other mutants FRET was significantlyreduced. The FRET change of mutants EcPutA1−1308 andA1309D/L1316D was <26% of wild-type indicating a severe lossin liposome binding. The lower changes in FRET with theEcPutA mutants are not a result of lower PRODH activity, as thePRODH activity of the mutants are similar to that of wild-typeEcPutA (Table 1). It is also important to note that the FADcofactors of all the mutants were fully reduced by the prolineconcentration used for the FRET experiments (data not shown).Thus, the FRET behavior of the EcPutA1−1308 and EcPutA

A1309D/L1316D mutants are consistent with the other resultsabove, altogether indicating that C-terminal residues 1309−1320have a critical role in EcPutA−membrane binding.

Role of CCM Residues in Functional Switching. Toexamine the effects of CCM truncations on the in vivotranscriptional repressor activity of EcPutA, we performed cell-based reporter assays. In agreement with our previousstudies,44,48 repression of the putC:lacZ reporter gene isattenuated upon proline addition as wild-type EcPutA binds tothe membrane rather than to put control DNA (Figure 6). Incontrast, repression of the reporter gene is not relieved by prolinein cells expressing EcPutA1−1314, EcPutA1−1308, or L1316D,providing further evidence that these mutants are unable toassociate with the membrane. Intriguingly, EcPutA A1309/

Figure 4. Lipid binding assays of EcPutA CCMmutants. (A) Lipid pull-down assays were performed with EcPutA wild-type and mutants (0.3mg/mL) in 10 mM HEPES buffer (pH 7.5, 150 mM NaCl) with E. colipolar lipids (0.8 mg/mL) for 1 h in the absence (Ox) and presence(Red) of proline (20 mM). The soluble and lipid fractions wereseparated by Air-fuge ultracentrifugation and analyzed by SDS-PAGE aspreviously described.48 (B) Protein bands from SDS-PAGE werequantified using ImageJ software. The percent distribution of EcPutA inthe soluble and lipid fractions is shown for the oxidized and prolinereduced samples as different colored bars (black, oxidized soluble; red,oxidized lipid; green, proline reduced soluble; yellow, proline reducedlipid). A partition coefficient calculated as (S/L)ox/(S/L)red aspreviously described48 is a measure of the extent that EcPutA partitionsinto the lipids in the presence of proline and is indicated by open circles.

Figure 5. FRET analysis of EcPutA−lipid binding. (A) Wild-typeEcPutA (1 μM)was incubated with 0.025mg/mL E. coli polar liposomes(containing 5% dansyl-PE) in 10 mM HEPES buffer (150 mM NaCl, 2mM MgCl2, pH 7.5) and titrated with 0−10 mM proline. The sampleswere excited at 290 nm (λex = 290 nm), and the emission spectra of theTrp residues and the dansyl liposomes were monitored at 340 and 520nm, respectively. With 10 mM proline, the decrease in Trp fluorescencewas 14.4%, and the corresponding increase in dansyl emission was12.7%. (B) EcPutA wild-type and mutants (1 μM) were incubated with0.025 mg/mL E. coli polar vesicles (containing 5% dansyl-PE) in 10 mMHEPES buffer (150 mMNaCl, 2 mMMgCl2, pH 7.5) with and without5mMproline. FRETwas monitored at λem = 520 nm using λex = 290 nm.Shown is the relative change in fluorescence emission of the dansylliposomes as a function of proline with the mutants normalized to wild-type EcPutA. Values are the mean of three independent experiments ±SD (*, p < 0.05).

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6297

L1316D acts as a “super-repressor”, where there is no discernibleexpression of the reporter in the absence or presence of proline.Thus, dysfunctional membrane binding impairs the ability of theEcPutA mutants to function as a proline-responsive transcrip-tional repressor.We then explored whether CCM mutants were defective in

proline-dependent conformational changes. Limited proteolysisof wild-type EcPutA in the absence and presence of prolinegenerates distinct proteolysis patterns for the oxidized andproline-reduced forms of EcPutA (Figure 7). In agreement withour previous studies, proline induces a change from a 135-kDaband (oxidized) to a 119-kDa band (proline-reduced) (Figure7).46,49 Limited proteolysis was then performed with the EcPutAmutants A1309D, A1309D/L1316D, and EcPutA1−1308(Figure 7). EcPutA A1309D mutant exhibited a distinct 119-kDa band with proline similar to that of wild-type. EcPutA1−

1308 and A1309D/L1316D, however, exhibited no change withproline in the chymotrypsin proteolysis pattern and instead hadonly a major 111-kDa band in the absence and presence ofproline. These results indicate that EcPutA mutants EcPutA1−1308 and A1309D/L1316D are unable to undergo a proline-dependent conformational change similar to that of wild-type.Thus, alteration of the CCM by point mutation or truncationinhibits conformational changes that are associated withmembrane-binding and redox-sensitive functional switching ofEcPutA.

Membrane Binding Regions of EcPutA.We next exploredother regions in PutA for possible membrane interactions byutilizing a FRET mapping approach. Residues in differentdomains were mutated to Trp, while existing Trp residues weremutated to Phe. If a mutated residue is within ∼24 Å (estimatedForster distance for the Trp/dansyl liposome pairing)60,61 of themembrane, the FRET signal is expected to change, thusindicating membrane interactions within that region. Theaddition of Trp in the membrane-binding domain shouldincrease FRET, whereas removal of Trp from the membrane-binding domain should decrease FRET. Substitutions at residuesfar from the PutA−membrane interaction surface would have noimpact on FRET.Residues in the α-domain and the CCM were the primary

focus of mutagenesis for the FRET study. Within the α-domain,residues Phe259, Phe205, and Met255 were mutated to Trp,while Trp194 and Trp211 were mutated to Phe. In the CCM, wemutated the terminal Gly1320 residue to Trp. In addition, wealso performed mutagenesis of residues in the auxiliary armdomain, as this domain is found within the Ser58−Asn233 regionimplicated in previous studies to be vital for membranebinding.46 The auxiliary arm contains no Trp residues; thereforewe introduced Trp at residues Tyr129 and Leu131. We alsomutated residue Trp1185 to Phe in the ALDHSF domain.Figure 8 shows that EcPutAmutantsW194F andW211F in the

α-domain exhibited a decrease in FRET change with prolinerelative to wild-type EcPutA. A double mutant of these residues(W194F/W211F) resulted in a more pronounced decrease inFRET change. These results suggest that the α-domain interactswith the membrane and that Trp194 and Trp211 are significant

Figure 6. Cell-based functional switching assays. Cell-based transcrip-tional repressor assays in E. coli strain JT31 putA− lacZ− cells containingthe putC:lacZ reporter gene plasmid and expression constructs forEcPutA wild-type and mutants EcPutA1−1314, EcPutA1−1308,L1316D, and A1309D/L1316D. Cells were grown at 37 °C in minimalmedia supplemented with 0 and 15mMproline. β-Galactosidase activityis reported as the average of four independent experiments ± standarderrors. BD, below detection.

Figure 7. Limited proteolysis of wild-type EcPutA and mutants A1309D, A1309D/L1316D, and EcPutA1−1308. Purified proteins (1 mg/mL) weredigested with chymotrypsin (10 μg/mL) for 1 h in 50 mM potassium phosphate (pH 7.5) alone (OX) and in reactions containing 20 mM proline(PRO). Digested protein fragments (10 μg) were separated by denaturing gel electrophoresis (7.5% acrylamide), and gels were stained with CoomassieBlue. The left-hand lane of each gel is undigested EcPutA protein (144 kDa band).

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6298

contributors to the FRET signal. To test the impact of insertingTrp residues in the α-domain, EcPutA mutants F259W, F205W,and M255W were also examined. All of these mutants generateda higher FRET signal change with proline, and double mutantsF259W/F205W and F259W/M255W led to further increases inFRET. These results strongly indicate that the α-domaininteracts with membranes.Adding a Trp residue to the CCM in EcPutA mutant G1320W

increased the proline-induced FRET change relative to wild-typeEcPutA. This result provides additional evidence for the CCMbeing involved in EcPutA−membrane binding. The PRODHdomain was also tested by replacing Trp396 with Phe. TheEcPutA mutant W396F exhibited a significant decrease in FRETwith proline, indicating that the PRODH domain associates withthe membrane, which is required for catalytic turnover.Mutations that did not significantly impact FRET were Y129Wand L131W of the arm domain and W1185F of the ALDHSFdomain. These results indicate that the regions surroundingthese residues do not associate with the membrane.To confirm that FRET changes observed with the mutants in

the α-domain and CCM are not caused by changes in PRODHactivity, the steady-state kinetic parameters for PRODH activitywere determined. Table 1 shows that all of the mutants in thisstudy exhibited PRODH activity similar to that of wild-typeEcPutA. We also confirmed that mutants exhibiting decreasedFRET are capable of functionally interacting with the membrane.Figure 9 shows that EcPutA mutants W396F, W194F, W211F,and W194F/W211F have functional membrane binding activitysimilar to that of the wild-type EcPutA, indicating that thedecrease in FRET observed with these mutants is not caused byimpaired membrane interactions.

■ DISCUSSION

The mechanism by which trifunctional PutAs functionally switchfrom a transcription factor to an active membrane-bound enzymeis currently unknown. In order to determine the mechanism, themembrane-binding domain of PutA needs to be established.Here, we located the regions of EcPutA involved in membrane

binding. We show that residues of the CCM are of crucialimportance for EcPutA functional membrane interactions.Truncation of EcPutA to residue 1314 attenuates its ability tofunctionally associate with membrane vesicles, and furtherdeletion to residue 1308 (EcPutA1−1308) completely abolishesmembrane-binding activity. Also, deleting the CCM in BjPutA(BjPutA1−986) eliminates TTC reduction in cell-based assays,suggesting that membrane binding is a conserved role of theCCM in all three PutA types. Crystal structures of type A andtype B PutAs show that the CCM is part of a lid that helps sealsthe substrate channel from the bulk medium.27−29,34,35 Thus, itappears that the CCM has a dual function in PutAs.The CCMwas further examined in EcPutA by replacing highly

conserved Ala1309 and Leu1316 with a negatively charged Asp

Figure 8. FRET mapping of EcPutA membrane interactions. Wild-type EcPutA and mutants (1 μM) were incubated with inverted E. coli polar vesicles(with 5% dansyl-PE incorporated) (0.025 mg/mL) in 10 mM HEPES buffer (150 mM NaCl, 2 mM MgCl2, pH 7.5) and treated with and without 10mMproline. Fluorescence emission was monitored at λem = 520 nm using a λex = 290 nm. (Top panel) The coloring of the EcPutA domain mapmatchesthe colors of the FRET data (bottom panel). FRET data are plotted as the relative change in fluorescence emission as a function of proline with themutants normalized to wild-type EcPutA. Data are the mean ± SD from four independent experiments (*, p < 0.05).

Figure 9. Functional membrane association assays. EcPutA wild-typeand mutant proteins (4 μg) were incubated with E. coli (strain JT31putA−) inverted membrane vesicles (0.1 mg/mL membrane protein) in20 mMMOPS (pH 7.5) containing 60 mM proline and 4 mM o-AB. Allreactions were performed at 23 °C, and absorbance was monitored at443 nm, which follows o-AB-P5C adduct formation. Plotted is theconcentration of o-AB-P5C complex formed per mg of EcPutA protein.

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6299

residue to potentially disrupt hydrophobic PutA−membraneinteractions. The EcPutA mutants A1309D and L1316D hadattenuated functional membrane association activity, while thedouble mutant A1309D/L1316D showed no activity similar tothat of EcPutA1−1308. Lipid pull-down assays and FRETexperiments also showed that substituting Asp at hydrophobicresidues in the CCM severely impairs EcPutA−membranebinding. These results are consistent with PutA−membranebinding involving hydrophobic interactions as Surber and Maloyoriginally proposed.62 Other studies have also concluded EcPutAand BjPutA membrane binding involves hydrophobic inter-actions. Isothermal titration calorimetry revealed BjPutA−membrane binding is entropy driven consistent with stronghydrophobic forces being involved.57 However, BjPutA alsoexhibits an electrostatic dependence on lipid binding, whereasEcPutA does not discriminate between differently chargedmembranes.45,57 From the results here, it seems that highlyconserved residues Ala1309 and Leu1316 of EcPutA have a keyrole in driving hydrophobic PutA−membrane binding.In addition to disrupting membrane binding, EcPutA mutants

EcPutA1−1308 and A1309D/L1316D were shown not toundergo a proline-dependent conformational change as observedfor wild-type EcPutA. In EcPutA, proline induces a conforma-tional change that is concomitant with PutA−membranebinding. The lack of a proline-induced conformational changein EcPutA mutants EcPutA1−1308 and A1309D/L1316D isconsistent with the loss of membrane binding. In vivo cell-basedassays showed that EcPutA1−1308 and A1309D/L1316D do notfunctionally switch. Interestingly, the A1309D/L1316D mutantappears to behave as a super-repressor of the put regulon.Besides the CCM, we also further demonstrated that the α-

domain and the PRODH domain are involved in EcPutAmembrane binding using a FRET mutagenesis mapping

approach. Dansyl liposomes are a useful tool for monitoringprotein−membrane interactions.60,63−66 Interactions betweenlipid vesicles and synaptotagmin upon calcium addition weremeasured using intrinsic tryptophan fluorescence from synapto-tagmin and dansyl liposomes.63 Synaptotagmin and dansyl-PEliposomes incubated in the presence of calcium showed a higherFRET response than synaptotagmin and dansyl liposomes alone.Dansyl lipids were also used as the acceptor in FRET assaysexamining the membrane binding capabilities of Factor VIII.65

Here, we utilized the tryptophan donor/dansyl acceptor systemto create a series of mapping mutants. By introducing orremoving Trp residues, we altered the Trp donor fluorescence,thereby manipulating dansyl fluorescence emission if themutation was at a residue near the membrane-binding site.Replacing Trp residues in the α-domain and PRODH domainwith Phe decreased FRET indicating that these Trp residues arewithin the membrane-binding region. Substituting in Trpresidues in the α-domain caused a FRET increase, furtherimplicating the α-domain in membrane binding and corroborat-ing with previous studies indicating a membrane-binding regionwithin residues 58−223.46 Adding a Trp residue in the CCM atGly1320 (G1320W) also increased FRET further verifying amembrane-binding role for the CCM. In contrast, adding Trpresidues within the arm domain or mutating Trp1185 tophenylalanine in the ALDHF domain did not change FRET,indicating these regions are not involved in membrane binding.In summary, our results provide strong evidence for the α-domain, PRODH domain, and the CCM as being critical forEcPutA membrane interactions.The residues examined in this study are highlighted in a

homology model of the catalytic module of EcPutA in Figure 10.The homologymodel was built using SWISS-MODEL67 with thecoordinates of the crystal structure of PutA (PDB entry 5KF6)

Figure 10. Structure-based summary of FRETmapping data. (A)Homology model of the catalytic module of EcPutA. Themodel includes residues 85−1316. Red spheres indicate residues that are within the Forster distance of the EcPutA−membrane interaction surface (positive FRET hits). Blue spheresindicate residues outside of the Forster distance of the EcPutA−membrane interaction surface (negative FRET hits). The white sphere indicates thetarget area obtained by trilateration of the positive FRET hits. (B) Zoom of the boxed area indicated in panel A, which contains the target area fromtrilateration. (C) Surface representation of the zoom region highlighting the convergence of three domains.

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6300

from Sinorhizobium meliloti (SmPutA) as the template.29

Although SmPutA is a type B PutA, it has 60% sequence identityto EcPutA and contains all of the domains being probed in theFRET experiments.29 The model contains residues 85−1316,which includes all but the DNA-binding domain.Although separated in primary sequence, themodel shows that

the α-domain and the CCMtwo regions implicated here inmembrane bindingconverge spatially near the FAD. Theresidues identified by FRET as near the membrane interactionsurface are in the CCM, α-domain, and the bottom of thePRODH domain (Trp396) (red spheres in Figure 10). Theintersection of 24-Å radius spheres centered on these residuesdefines a target that is consistent with all the FRET data andpresumably within the membrane-association region (whitesphere in Figure 10). Importantly, the trilateration target isoutside of the 24-Å radii spheres centered on the FRET-negativeresidues (blue spheres in Figure 10). This calculation predicts ajunction formed by the CCM, α-domain, PRODH domain, andGSALDH domain as potentially involved in membrane binding(Figure 10B,C). As mentioned above, the α-domain is also thebinding site for a detergent molecule observed in a GsPutAstructure (PDB entry 4NMC). Figure 10A shows the location ofthe detergent-binding site fromGsPutA in the EcPutA homologymodel. Interestingly, this region includes helix α8 of the PRODHactive site, which has been shown to undergo conformationalchanges in response to proline binding and flavin reduction.28,49

Because of its flexibility, helix α8 has been suggested to helptransmit the flavin redox state to the membrane-binding domain.The model in Figure 10 is consistent with this notion. Also, theCCMbeing adjacent to the α-domain and helix α8 provides cluesinto why deletion of the CCM may impair proline-inducedconformational changes and membrane binding of EcPutA.Now that a membrane-binding interface of EcPutA has been

identified, it may be possible to map a hydrogen bond networkbetween the FAD cofactor and the membrane-binding interface.We propose that when the FAD is reduced by proline,conformational changes in EcPutA enable the α-domain andCCM to interact with the membrane. Determining themechanism of this process will generate novel insights intoflavin-based redox functional switching and may be useful forinhibiting proline metabolism in pathogenic organisms thatutilize trifunctional PutAs.

■ ASSOCIATED CONTENT

*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.bio-chem.7b01008.

Tables S1 and S2, and Figure S1 (PDF)

■ AUTHOR INFORMATION

Corresponding Author*E-mail: dbecker3@unl.edu. Phone: 402-472-9652. Fax: 402-472-472-7842.

ORCID

John J. Tanner: 0000-0001-8314-113XDonald F. Becker: 0000-0002-1350-7201Author Contributions‡S.L.C. and W.Z. contributed equally to this work.

FundingThis research was supported by NIH Grants GM061068 andP30GM103335, NSF Grant DBI-1156692, and by the Universityof Nebraska Agricultural Research Division, supported in part byfunds provided through the Hatch Act.NotesThe authors declare no competing financial interest.

■ ABBREVIATIONSALDHSF, aldehyde dehydrogenase superfamily; CCM, C-terminal motif; dansyl-PE, 1,2-dioleoyl-sn-glycero-3-phosphoe-thanolamine-N-(5-dimethylamino-1-naphthalenesulfonyl;DCPIP, dichlorophenolindophenol; BjPutA, proline utilizationA from Bradyrhizobium japonicum; EcPutA, proline utilization Afrom Escherichia coli; FAD, flavin adenine dinucleotide; FRET,Forster resonance energy transfer; GsPutA, proline utilization AfromGeobacter sulfurreducens; GSAL, glutamate-γ-semialdehyde;GSALDH, GSAL dehydrogenase; HEPES, N-(2-hydroxyethyl)-piperazine-N′-2-ethanesulfonic acid; MOPS, 3-(N-morpholino)-propanesulfonic acid; NAD+, nicotinamide adenine dinucleo-tide; o-AB, o-aminobenzaldehyde; PE, phosphoethanolamine;PG, phosphatidylglycerol; PRODH, proline dehydrogenase;P5C, Δ1-pyrroline-5-carboxylate; PutA, proline utilization A;RcPutA, proline utilization A from Rhodobacter capsulatus; RHH,ribbon−helix−helix; SDS-PAGE, sodium dodecyl sulfate poly-acrylamide gel electrophoresis; SmPutA, proline utilization Afrom Sinorhizobium meliloti; TTC, triphenyltetrazolium chloride

■ REFERENCES(1) Szabados, L., and Savoure, A. (2010) Proline: a multifunctionalamino acid. Trends Plant Sci. 15, 89−97.(2) Phang, J. M. (2017) Proline metabolism in cell regulation andcancer biology: Recent advances and hypotheses. Antioxid. RedoxSignaling. DOI: 10.1089/ars.2017.7350.(3) Arentson, B. W., Sanyal, N., and Becker, D. F. (2012) Substratechanneling in proline metabolism. Front. Biosci., Landmark Ed. 17, 375−388.(4) Phang, J. M., Donald, S. P., Pandhare, J., and Liu, Y. (2008) Themetabolism of proline, a stress substrate, modulates carcinogenicpathways. Amino Acids 35, 681−690.(5) Liu, L. K., Becker, D. F., and Tanner, J. J. (2017) Structure,function, and mechanism of proline utilization A (PutA). Arch. Biochem.Biophys. 632, 142−157.(6) Adams, E., and Frank, L. (1980) Metabolism of proline and thehydroxyprolines. Annu. Rev. Biochem. 49, 1005−1061.(7) Liu, W., Le, A., Hancock, C., Lane, A. N., Dang, C. V., Fan, T. W.,and Phang, J. M. (2012) Reprogramming of proline and glutaminemetabolism contributes to the proliferative and metabolic responsesregulated by oncogenic transcription factor c-MYC. Proc. Natl. Acad. Sci.U. S. A. 109, 8983−8988.(8) Phang, J. M., Liu, W., Hancock, C. N., and Fischer, J. W. (2015)Proline metabolism and cancer: emerging links to glutamine andcollagen. Curr. Opin. Clin. Nutr. Metab. Care 18, 71−77.(9) Elia, I., Broekaert, D., Christen, S., Boon, R., Radaelli, E., Orth, M.F., Verfaillie, C., Grunewald, T. G. P., and Fendt, S. M. (2017) Prolinemetabolism supports metastasis formation and could be inhibited toselectively target metastasizing cancer cells. Nat. Commun. 8, 15267.(10) Olivares, O., Mayers, J. R., Gouirand, V., Torrence, M. E., Gicquel,T., Borge, L., Lac, S., Roques, J., Lavaut, M. N., Berthezene, P., Rubis, M.,Secq, V., Garcia, S., Moutardier, V., Lombardo, D., Iovanna, J. L.,Tomasini, R., Guillaumond, F., Vander Heiden, M. G., and Vasseur, S.(2017) Collagen-derived proline promotes pancreatic ductal adeno-carcinoma cell survival under nutrient limited conditions.Nat. Commun.8, 16031.(11) Clelland, C. L., Read, L. L., Baraldi, A. N., Bart, C. P., Pappas, C. A.,Panek, L. J., Nadrich, R. H., and Clelland, J. D. (2011) Evidence for

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6301

association of hyperprolinemia with schizophrenia and a measure ofclinical outcome. Schizophr Res. 131, 139−145.(12) Willis, A., Bender, H. U., Steel, G., and Valle, D. (2008) PRODHvariants and risk for schizophrenia. Amino Acids 35, 673−679.(13) Zarchi, O., Carmel, M., Avni, C., Attias, J., Frisch, A.,Michaelovsky, E., Patya, M., Green, T., Weinberger, R., Weizman, A.,and Gothelf, D. (2013) Schizophrenia-like neurophysiological abnor-malities in 22q11.2 deletion syndrome and their association to COMTand PRODH genotypes. J. Psychiatr. Res. 47, 1623−1629.(14) Radoeva, P. D., Coman, I. L., Salazar, C. A., Gentile, K. L., Higgins,A. M., Middleton, F. A., Antshel, K. M., Fremont, W., Shprintzen, R. J.,Morrow, B. E., and Kates, W. R. (2014) Association between autismspectrum disorder in individuals with velocardiofacial (22q11.2deletion) syndrome and PRODH and COMT genotypes. Psychiatr.Genet. 24, 269−272.(15) Halsey, C. R., Lei, S., Wax, J. K., Lehman, M. K., Nuxoll, A. S.,Steinke, L., Sadykov, M., Powers, R., and Fey, P. D. (2017) Amino AcidCatabolism in Staphylococcus aureus and the Function of CarbonCatabolite Repression. mBio 8, No. e01434-16.(16) Schwan, W. R., Coulter, S. N., Ng, E. Y., Langhorne, M. H.,Ritchie, H. D., Brody, L. L., Westbrock-Wadman, S., Bayer, A. S., Folger,K. R., and Stover, C. K. (1998) Identification and characterization of thePutP proline permease that contributes to in vivo survival ofStaphylococcus aureus in animal models. Infect. Immun. 66, 567−572.(17) Krishnan, N., Doster, A. R., Duhamel, G. E., and Becker, D. F.(2008) Characterization of a Helicobacter hepaticus putA mutant strainin host colonization and oxidative stress. Infect. Immun. 76, 3037−3044.(18) Nakajima, K., Inatsu, S., Mizote, T., Nagata, Y., Aoyama, K.,Fukuda, Y., and Nagata, K. (2008) Possible involvement of put A gene inHelicobacter pylori colonization in the stomach and motility. Biomed.Res. 29, 9−18.(19) Lee, I. R., Lui, E. Y., Chow, E. W., Arras, S. D., Morrow, C. A., andFraser, J. A. (2013) Reactive oxygen species homeostasis and virulenceof the fungal pathogen Cryptococcus neoformans requires an intactproline catabolism pathway. Genetics 194, 421−433.(20) Paes, L. S., Suarez Mantilla, B., Zimbres, F. M., Pral, E. M., Diogode Melo, P., Tahara, E. B., Kowaltowski, A. J., Elias, M. C., and Silber, A.M. (2013) Proline dehydrogenase regulates redox state and respiratorymetabolism in Trypanosoma cruzi. PLoS One 8, e69419.(21) Bringaud, F., Riviere, L., and Coustou, V. (2006) Energymetabolism of trypanosomatids: adaptation to available carbon sources.Mol. Biochem. Parasitol. 149, 1−9.(22) Mantilla, B. S., Marchese, L., Casas-Sanchez, A., Dyer, N. A., Ejeh,N., Biran, M., Bringaud, F., Lehane, M. J., Acosta-Serrano, A., and Silber,A. M. (2017) Proline metabolism is essential for Trypanosoma bruceisurvival in the tsetse vector. PLoS Pathog. 13, e1006158.(23) Pemberton, T. A., Srivastava, D., Sanyal, N., Henzl, M. T., Becker,D. F., and Tanner, J. J. (2014) Structural studies of yeast Delta(1)-pyrroline-5-carboxylate dehydrogenase (ALDH4A1): active site flexi-bility and oligomeric state. Biochemistry 53, 1350−1359.(24)Moxley, M. A., and Becker, D. F. (2012) Rapid reaction kinetics ofproline dehydrogenase in the multifunctional proline utilization Aprotein. Biochemistry 51, 511−520.(25) Moxley, M. A., Tanner, J. J., and Becker, D. F. (2011) Steady-statekinetic mechanism of the proline:ubiquinone oxidoreductase activity ofproline utilization A (PutA) from Escherichia coli. Arch. Biochem.Biophys. 516, 113−120.(26) Wanduragala, S., Sanyal, N., Liang, X., and Becker, D. F. (2010)Purification and characterization of Put1p from Saccharomycescerevisiae. Arch. Biochem. Biophys. 498, 136−142.(27) Srivastava, D., Schuermann, J. P., White, T. A., Krishnan, N.,Sanyal, N., Hura, G. L., Tan, A., Henzl, M. T., Becker, D. F., and Tanner,J. J. (2010) Crystal structure of the bifunctional proline utilization Aflavoenzyme from Bradyrhizobium japonicum. Proc. Natl. Acad. Sci. U. S.A. 107, 2878−2883.(28) Singh, H., Arentson, B. W., Becker, D. F., and Tanner, J. J. (2014)Structures of the PutA peripheral membrane flavoenzyme reveal adynamic substrate-channeling tunnel and the quinone-binding site. Proc.Natl. Acad. Sci. U. S. A. 111, 3389−3394.

(29) Luo, M., Gamage, T. T., Arentson, B. W., Schlasner, K. N., Becker,D. F., and Tanner, J. J. (2016) Structures of Proline Utilization A (PutA)Reveal the Fold and Functions of the Aldehyde DehydrogenaseSuperfamily Domain of Unknown Function. J. Biol. Chem. 291, 24065−24075.(30) Moxley, M. A., Sanyal, N., Krishnan, N., Tanner, J. J., and Becker,D. F. (2014) Evidence for hysteretic substrate channeling in the prolinedehydrogenase and Delta1-pyrroline-5-carboxylate dehydrogenasecoupled reaction of proline utilization A (PutA). J. Biol. Chem. 289,3639−3651.(31) Arentson, B. W., Luo, M., Pemberton, T. A., Tanner, J. J., andBecker, D. F. (2014) Kinetic and structural characterization of tunnel-perturbing mutants in Bradyrhizobium japonicum proline utilization A.Biochemistry 53, 5150−5161.(32) Surber, M. W., and Maloy, S. (1998) The PutA protein ofSalmonella typhimurium catalyzes the two steps of proline degradationvia a leaky channel. Arch. Biochem. Biophys. 354, 281−287.(33) Sanyal, N., Arentson, B. W., Luo, M., Tanner, J. J., and Becker, D.F. (2015) First evidence for substrate channeling between prolinecatabolic enzymes: a validation of domain fusion analysis for predictingprotein-protein interactions. J. Biol. Chem. 290, 2225−2234.(34) Korasick, D. A., Singh, H., Pemberton, T. A., Luo, M., Dhatwalia,R., and Tanner, J. J. (2017) Biophysical investigation of type A PutAsreveals a conserved core oligomeric structure. FEBS J. 284, 3029−3049.(35) Korasick, D. A., Gamage, T. T., Christgen, S., Stiers, K. M.,Beamer, L. J., Henzl, M. T., Becker, D. F., and Tanner, J. J. (2017)Structure and characterization of a class 3B proline utilization A: Ligand-induced dimerization and importance of the C-terminal domain forcatalysis. J. Biol. Chem. 292, 9652−9665.(36) Luo, M., Christgen, S., Sanyal, N., Arentson, B. W., Becker, D. F.,and Tanner, J. J. (2014) Evidence that the C-terminal domain of a type BPutA protein contributes to aldehyde dehydrogenase activity andsubstrate channeling. Biochemistry 53, 5661−5673.(37) Zhou, Y., Larson, J. D., Bottoms, C. A., Arturo, E. C., Henzl, M. T.,Jenkins, J. L., Nix, J. C., Becker, D. F., and Tanner, J. J. (2008) Structuralbasis of the transcriptional regulation of the proline utilization regulonby multifunctional PutA. J. Mol. Biol. 381, 174−188.(38) Larson, J. D., Jenkins, J. L., Schuermann, J. P., Zhou, Y., Becker, D.F., and Tanner, J. J. (2006) Crystal structures of the DNA-bindingdomain of Escherichia coli proline utilization A flavoprotein and analysisof the role of Lys9 in DNA recognition. Protein Sci. 15, 2630−2641.(39) Halouska, S., Zhou, Y., Becker, D. F., and Powers, R. (2009)Solution structure of the Pseudomonas putida protein PpPutA45 and itsDNA complex. Proteins: Struct., Funct., Genet. 75, 12−27.(40) Bracher, S., Schmidt, C. C., Dittmer, S. I., and Jung, H. (2016)Core Transmembrane Domain 6 Plays a Pivotal Role in the TransportCycle of the Sodium/Proline Symporter PutP. J. Biol. Chem. 291,26208−26215.(41) Arentson, B. W., Hayes, E. L., Zhu, W., Singh, H., Tanner, J. J., andBecker, D. F. (2016) Engineering a trifunctional proline utilization Achimaera by fusing a DNA-binding domain to a bifunctional PutA. Biosci.Rep. 36, No. e00413.(42) Becker, D. F., Zhu, W., and Moxley, M. A. (2011) Flavin redoxswitching of protein functions. Antioxid. Redox Signaling 14, 1079−1091.(43) Ostrovsky de Spicer, P., and Maloy, S. (1993) PutA protein, amembrane-associated flavin dehydrogenase, acts as a redox-dependenttranscriptional regulator. Proc. Natl. Acad. Sci. U. S. A. 90, 4295−4298.(44) Zhou, Y., Zhu, W., Bellur, P. S., Rewinkel, D., and Becker, D. F.(2008) Direct linking of metabolism and gene expression in the prolineutilization A protein from Escherichia coli. Amino Acids 35, 711−718.(45) Zhang, W., Zhou, Y., and Becker, D. F. (2004) Regulation ofPutA-membrane associations by flavin adenine dinucleotide reduction.Biochemistry 43, 13165−13174.(46) Zhu, W., and Becker, D. F. (2003) Flavin redox state triggersconformational changes in the PutA protein from Escherichia coli.Biochemistry 42, 5469−5477.(47) Brown, E. D., and Wood, J. M. (1993) Conformational changeand membrane association of the PutA protein are coincident withreduction of its FAD cofactor by proline. J. Biol. Chem. 268, 8972−8979.

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6302

(48) Zhu, W., Haile, A. M., Singh, R. K., Larson, J. D., Smithen, D.,Chan, J. Y., Tanner, J. J., and Becker, D. F. (2013) Involvement of thebeta3-alpha3 loop of the proline dehydrogenase domain in allostericregulation of membrane association of proline utilization A. Biochemistry52, 4482−4491.(49) Srivastava, D., Zhu, W., Johnson, W. H., Jr., Whitman, C. P.,Becker, D. F., and Tanner, J. J. (2010) The structure of the prolineutilization a proline dehydrogenase domain inactivated by N-propargylglycine provides insight into conformational changes inducedby substrate binding and flavin reduction. Biochemistry 49, 560−569.(50) Zhang, W., Zhang, M., Zhu, W., Zhou, Y., Wanduragala, S.,Rewinkel, D., Tanner, J. J., and Becker, D. F. (2007) Redox-inducedchanges in flavin structure and roles of flavin N(5) and the ribityl 2′-OHgroup in regulating PutA-membrane binding. Biochemistry 46, 483−491.(51) Lee, Y. H., Nadaraia, S., Gu, D., Becker, D. F., and Tanner, J. J.(2003) Structure of the proline dehydrogenase domain of themultifunctional PutA flavoprotein. Nat. Struct. Biol. 10, 109−114.(52) Zhu, W., and Becker, D. F. (2005) Exploring the proline-dependent conformational change in the multifunctional PutAflavoprotein by tryptophan fluorescence spectroscopy. Biochemistry 44,12297−12306.(53) Moxley, M. A., Zhang, L., Christgen, S., Tanner, J. J., and Becker,D. F. (2017) Identification of a conserved histidine as being critical forthe catalytic mechanism and functional switching of the multifunctionalproline utilization A protein. Biochemistry 56, 3078−3088.(54) Abrahamson, J. L., Baker, L. G., Stephenson, J. T., andWood, J. M.(1983) Proline dehydrogenase from Escherichia coli K12. Properties ofthe membrane-associated enzyme. Eur. J. Biochem. 134, 77−82.(55)Miller, J. H. (1972) Experiments inMolecular Genetics, Cold SpringHarbor Laboratory, Cold Spring Harbor, NY.(56)Wood, J. M. (1981) Genetics of L-proline utilization in Escherichiacoli. J. Bacteriol. 146, 895−901.(57) Zhang, W., Krishnan, N., and Becker, D. F. (2006) Kinetic andthermodynamic analysis of Bradyrhizobium japonicum PutA-membraneassociations. Arch. Biochem. Biophys. 445, 174−183.(58) Berridge, M. V., Herst, P. M., and Tan, A. S. (2005) Tetrazoliumdyes as tools in cell biology: new insights into their cellular reduction.Biotechnol. Annu. Rev. 11, 127−152.(59) Rich, P. R., Mischis, L. A., Purton, S., and Wiskich, J. T. (2001)The sites of interaction of triphenyltetrazolium chloride withmitochondrial respiratory chains. FEMS Microbiol. Lett. 202, 181−187.(60) Loura, L. M., Prieto, M., and Fernandes, F. (2010) Quantificationof protein-lipid selectivity using FRET. Eur. Biophys. J. 39, 565−578.(61) Wang, S., Martin, E., Cimino, J., Omann, G., and Glaser, M.(1988) Distribution of phospholipids around gramicidin and D-beta-hydroxybutyrate dehydrogenase as measured by resonance energytransfer. Biochemistry 27, 2033−2039.(62) Surber, M. W., and Maloy, S. (1999) Regulation of flavindehydrogenase compartmentalization: requirements for PutA-mem-brane association in Salmonella typhimurium. Biochim. Biophys. Acta,Biomembr. 1421, 5−18.(63) Brose, N., Petrenko, A. G., Sudhof, T. C., and Jahn, R. (1992)Synaptotagmin: a calcium sensor on the synaptic vesicle surface. Science256, 1021−1025.(64) Nalefski, E. A., Sultzman, L. A., Martin, D. M., Kriz, R. W., Towler,P. S., Knopf, J. L., and Clark, J. D. (1994) Delineation of two functionallydistinct domains of cytosolic phospholipase A2, a regulatory Ca(2+)-dependent lipid-binding domain and a Ca(2+)-independent catalyticdomain. J. Biol. Chem. 269, 18239−18249.(65) Bardelle, C., Furie, B., Furie, B. C., and Gilbert, G. E. (1993)Membrane binding kinetics of factor VIII indicate a complex bindingprocess. J. Biol. Chem. 268, 8815−8824.(66) Nalefski, E. A., and Newton, A. C. (2001) Membrane bindingkinetics of protein kinase C betaII mediated by the C2 domain.Biochemistry 40, 13216−13229.(67) Biasini, M., Bienert, S., Waterhouse, A., Arnold, K., Studer, G.,Schmidt, T., Kiefer, F., Gallo Cassarino, T., Bertoni, M., Bordoli, L., andSchwede, T. (2014) SWISS-MODEL: modelling protein tertiary and

quaternary structure using evolutionary information. Nucleic Acids Res.42, W252−W258.

Biochemistry Article

DOI: 10.1021/acs.biochem.7b01008Biochemistry 2017, 56, 6292−6303

6303